Dynamic Specifications for Sampling A D Converters 1 0 INTRODUCTION Traditionally analog-to-digital converters (ADCs) have been specified by their static characteristics such as integral and differential nonlinearity gain error and offset error These specifications are important for determining the DC accuracy of an A D converter and are very important in applications such as weighing temperature measurement and other situations where the input signal varies slowly over time Many applications however require digitizing a signal which varies quickly over time These include digital signal processing (DSP) applications such as digital audio spectral analysis and motion control For these applications DC accuracy is not as crucial as AC accuracy The important specifications for these applications are the dynamic specifications such as signal-to-noise ratio total harmonic distortion intermodulation distortion and input bandwidth An A D converter by itself cannot accurately digitize high frequency signals due to limitations imposed by its conversion time (If the signal varies by more than LSB during the conversion time of the ADC the conversion will not be fully accurate ) To digitize a high frequency signal one must use a sample-and-hold (S H) amplifier to freeze the signal long enough so that the ADC can make an accurate conversion Hence the dynamic specifications are not unique to the ADC but are properties of the S H-ADC system With the advent of sampling ADCs which have a S H built onto the chip with the ADC it is now possible to meaningfully characterize the dynamic specifications of a single device In this application note we discuss the meaning and significance of the various dynamic specifications for sampling ADCs and in the appendix we give a table which has typical values for these specifications for a number of sampling ADCs 2 0 MEANING OF THE DYNAMIC SPECIFICATIONS Signal-to-Noise Ratio The signal-to-noise (S N) ratio is the ratio of the signal amplitude to the noise level It is National Semiconductor Application Note 769 Leon G Melkonian May 1991 generally specified in the data sheets at a set of input signal frequencies at a specific sampling rate and with the signal amplitude at or near the maximum allowable level There is some ambiguity regarding the composition of the noise component of the S N ratio Some manufacturers reserve the term S N ratio to include only the background noise and the spurious noise whereas others also include the harmonics of the signal When comparing the S N ratio for several sampling ADCs one should look in the data sheets to see how it is measured When the harmonics are included the S N specification is frequently referred to as the Signal-to-(Noise a Distortion) or SINAD Both signal-to-noise specifications exclude any DC offset from the noise component To determine the background noise level one must integrate the noise spectral density over the bandwidth of interest Even a perfect ADC will have some noise which arises from the quantization process If one treats this quantization noise as white noise and considers no other noise sources the maximum S N ratio attainable for an n-bit ADC is 1 S N e 6 02n a 1 76 db Hence one can see that the S N ratio can be increased by going to the higher resolution ADCs Total Harmonic Distortion or THD relates the rms sum of the amplitudes of the digitized signal s harmonics to the amplitude of the signal THD e V 2 f2 a V 2 f3 a V 2 f1 J where V f1 is the amplitude of the fundamental and V fi is the amplitude of the i th harmonic One generally includes all harmonics within the bandwidth of interest however sometimes in practice only the first five harmonics are taken into account because higher order harmonics have a negligible effect on the THD FIGURE 1 A Signal with a Single Frequency Component (left) Suffers Harmonic Distortion after A D Conversion (right) TL H 11193 1 Dynamic Specifications for Sampling A D Converters AN-769 C1995 National Semiconductor Corporation TL H 11193 RRD-B30M75 Printed in U S A
ADCs produce harmonics of an input signal because an ADC is an inherently nonlinear device This can be easily seen by looking at the transfer curve of an ideal ADC which looks like a staircase with equal-sized steps (Figure 2) Ina real ADC bowing and other nonlinearities add to the distortion If the output of an ADC is fed to a perfect DAC then the transfer function of this system can be represented in principle as a polynomial V OUT e a 0 a a 1 (V IN ) a a 2 (V IN )2 a a 3 (V IN )3 a A perfectly linear system would have all the a j zero except for a 0 and a 1 The second harmonic appears for example because a 2 is nonzero If one uses the trigonometric identity (cos 0t) 2 e 1 a cos 20t 2 one can see how a second-order nonlinearity produces an output that has a frequency that is twice that of the fundamental As one goes to higher resolution converters the THD will decrease because the transfer curve of the ADC more closely resembles a straight line (We are assuming that the TL H 11193 2 FIGURE 2 The Nonlinear Nature of the ADC Transfer Curve is the Cause of THD and IMD integral and differential nonlinearities are going down proportionately as the resolution increases This is generally the case ) Like the S N ratio the THD is generally specified in the data sheets at a set of frequencies and with the signal amplitude at or near the maximum allowable level It is usually specified in decibels or as a percentage Having a low THD is especially important in applications such as audio and spectral analysis because in these applications particularly one does not want the conversion process to add new frequency components to the signal Intermodulation Distortion or IMD results when two frequency components in a signal interact through the nonlinearities in the ADC to produce signals at additional frequencies If f a and f b are the signal frequencies at the input of the device the possible IMD products f mn are given by f mn e m f a g n f b where m and n take on positive integer values and are such that f mn is positive The j th order IMD products are those for which m a n e j The second order intermodulation products of f a and f b which occur when m and n both equal 1 are given by the sum and difference frequencies of f a and f b Using the formula (cos 0 1 t a cos 0 2 t) 2 e cos2 0 1 t a 2(cos 0 1 t)(cos 0 2 t) a cos20 2 t and the trigonometric identity 2(cos 0 1 t)(cos 0 2 t) e cos (0 1 a 0 2 )t a cos (0 1 b 0 2 )t one can see how a second-order nonlinearity leads to the production of an output that has components at the sum and difference frequencies of the inputs (and also at the second harmonics) The intermodulation distortion due to second order terms is commonly defined as IMD e V 2 1 1 a V 2 1 b1 V 2 a a V2 b J where V a and V b are the amplitudes of the fundamentals and V 1 1 and V 1 b1 are the amplitudes of the sum and difference frequencies respectively Figure 3 shows a particularly bad case of intermodulation distortion harmonic distortion products are also visible In calculating the IMD for this example one must include the higher order IMD products to get an accurate measure of the distortion FIGURE 3 An Input Signal with Frequency Components at 600 Hz and 1 khz (left) Suffers Severe IMD after A D Conversion (right) TL H 11193 3 2
The way IMD is specified for sampling ADCs varies from manufacturer to manufacturer Variations occur in the number of IMD products that are included in the measurement whether the two input frequencies have equal or unequal amplitudes and whether the distortion is referenced to the rms sum of the input amplitudes or to the amplitude of the larger input When comparing the IMD for several sampling ADCs especially if they are from different manufacturers one must know how it is measured in each case As one can infer from Figure 3 an ADC with a poor IMD specification would lead to very poor performance in an audio or spectral analysis application Another reason for wanting a good IMD specification is to prevent the appearance of spurious signals produced by the coupling of strong out-of-band frequency components with signals in the band of interest Peak Harmonic is the amplitude relative to the fundamental of the largest harmonic resulting from the A D conversion of a signal The peak harmonic is usually but not always the second harmonic It is usually specified in decibels Peak Harmonic or Spurious Noise is the amplitude relative to the signal level of the next largest frequency component (other than DC) Spurious noise components are noise components which are not integral multiples of the input signal and are often aliased harmonics if the signal frequency is a significant fraction of the sampling rate This specification is important because some applications require that the harmonics and spurious noise components be smaller than the lowest amplitude signal of interest Spurious-Free Dynamic Range is the ratio of the signal amplitude to the amplitude of the highest harmonic or spurious noise component the input signal amplitude is at or near full scale (Figure 4) This specification is simply the reciprocal of the Peak harmonic or spurious noise if in the measurement of that specification the input signal amplitude is at full scale Dynamic Differential Nonlinearity is the differential nonlinearity of the ADC for an AC input It is frequently measured at the maximum sampling rate and with an input that is at the Nyquist frequency (half the sampling rate) Differential nonlinearity is the deviation from the ideal 1 LSB input voltage span that is associated with each output code (Figure 5) It is measured using the histogram test TL H 11193 5 FIGURE 5 Differential Nonlinearity is a Measure of the Deviation from the Ideal 1 LSB Input Voltage Span Associated with Each Output Code Dynamic Integral Nonlinearity is the integral nonlinearity of the ADC for an AC input Like the dynamic differential nonlinearity it is frequently measured at Nyquist operation Integral nonlinearity describes the departure of the ADC transfer curve from the ideal transfer curve excluding the effects of offset gain and quantization errors (Figure 6) TL H 11193 4 FIGURE 4 Spurious-Free Dynamic Range Indicates How Far below Full Scale One Can Distinguish Signals from Noise and Distortion TL H 11193 6 FIGURE 6 Integral Nonlinearity Measures such Features as Bowing in an ADC Transfer Curve 3
Effective Number of Bits (ENOB) is a specification that is closely related to the signal-to-noise ratio It is determined by measuring the S N and using the equation S N b1 76 ENOB e 6 02 Some manufacturers define the effective number of bits using the SINAD instead of the signal-to-noise ratio The ENOB generally decreases at high frequencies and one frequently sees it plotted as a function of frequency along with the SINAD (Figure 7) The effective number of bits specification combines the effects of many of the other dynamic specifications Errors resulting from dynamic differential and integral nonlinearity missing codes total harmonic distortion and aperture jitter show up in the effective number of bits specification Full Power Bandwidth has several definitions A common definition is that it is the frequency at which the S N ratio has dropped by 3 db (relative to its low frequency level) for an input signal that is at or near the maximum allowable level This corresponds to a drop in the ENOB by bit relative to its low-frequency level Another definition is that it is the frequency at which the input signal appears to have been attenuated by 3 db Some manufacturers of flash converters define full power bandwidth as the frequency at which spurious or missing codes begin to appear (Missing codes will occur if the dynamic differential nonlinearity is greater than a1 LSB ) Small Signal Bandwidth is the frequency at which the S N ratio has dropped by 3 db for an input signal that is much smaller than the full-scale input 20 db or 40 db below fullscale for example The small signal bandwidth is generally larger than the full power bandwidth This will be the case if the bandwidth is slew rate limited for example The small signal bandwidth is important for those applications which do not require the conversion of large amplitude high frequency signals Relying on the full power bandwidth specification in these cases would constrain one to a smaller bandwidth than can actually be attained Sampling Rate or throughput rate depends on the length of the conversion time acquisition time and other time delays associated with carrying out a conversion Signals which have frequencies exceeding the Nyquist frequency (half the sampling rate) will be aliased to frequencies below the Nyquist frequency In order to prevent this signal degradation one must sample the signal at a rate that is more than twice the highest frequency component in the signal and or process the signal through a low pass (anti-aliasing filter) before it reaches the ADC In some applications one wants to sample at a rate much higher than the highest frequency component of interest in order to reduce the complexity of the anti-aliasing filter that is required SAMPLE-AND-HOLD CIRCUITRY SPECIFICATIONS Aquisition time aperture time and aperture jitter are specifications that relate to the internal sampling circuitry within the ADC These specifications can be easily explained with the aid of the simple sample-and-hold circuit shown in Figure 8 The S H amplifier is in Sample mode when the switch is closed In this case the output of the amplifier is following the input When the switch is opened the S H amplifier is in Hold mode in this case the output retains the input voltage that was present before the switch was opened FIGURE 7 Effective Number of Bits and S N Generally Drop at High Frequencies TL H 11193 7 4
TL H 11193 8 FIGURE 8 Simple S H Circuit Acquisition Time is the maximum time required to acquire a new input voltage once a Sample command has been given (Figure 9) A signal is acquired when it has settled to within a specified tolerance usually LSB of the input voltage The maximum value of the acquisition time occurs when the hold capacitor must charge up from zero to its fullscale value (or the other way around if it is larger) Acquisition time is important because it makes a significant contribution to the total time required to make a conversion TL H 11193 9 FIGURE 9 Acquisition Time Aperture Time is another specification that is defined differently by different manufacturers The strict definition is that it is the time during which the signal is being disconnected from the hold capacitor after a Hold command has been given The broader definition is that it is the time between the application of the Hold command and when the signal has been completely disconnected from the hold capacitor The second definition includes the digital delay which occurs between when the Hold command is applied and when the switch connecting the input signal to the hold capacitor begins to open Aperture time is important if one needs to acquire the value of a signal at a precise time Since the signal is not held instantaneously upon application of the Hold command this command must be given roughly an aperture time before one wants the signal to be frozen The aperture time is not a limiting factor on the frequency for sinusoidal signals because for a sinusoidal signal the voltage error caused by the aperture time manifests itself as a phase change not an amplitude or frequency change Aperture Jitter is the uncertainty in the aperture time Aperture jitter results from noise which is superimposed on the Hold command which affects its timing Aperture jitter is generally specified as an rms value which represents the standard deviation in the aperture time The aperture jitter sets an upper limit on the maximum frequency sinusoidal signal that can be accurately converted In order not to lose accuracy the rule of thumb is that the signal must not change by more than g LSB during the aperture jitter time Using a full-scale sinusoidal signal V e A sin (2qft) we have dv dt e 2qfA cos (2qft) g LSB k t aj where t aj is the aperture jitter Since LSB e A 2n where n is the resolution of the converter we get 1 f k 2q 2n t aj As an example of using lthis criterion a 12-bit converter whose S H amplifier has an aperture jitter of 100 ps could convert full-scale signals having frequencies as high as 388 khz Of course this would only be possible if the converter s sampling rate is at least twice as high as this frequency in order to satisfy the Nyquist criterion 3 0 CONCLUSION It is important to have a working knowledge of the dynamic specifications of sampling ADCs in order to be able to select a converter that is suitable for a specific system need One must also be aware that the test conditions and even the definitions of the specifications may vary from manufacturer to manufacturer One must take these variations into account when comparing ADCs Using the information in this note one should be able to understand the meanings of the various specifications and determine which ones are important for the particular application at hand REFERENCE 1 B Blesser Digitization of Audio J Audio Eng Soc vol 26 no 10 pp 742-743 (1978) 5
APPENDIX What follows is a table which presents values of the dynamic specifications for a number of National s sampling ADCs Additional information such as the values of S N and THD at frequencies not listed here can be found in the data sheets The reference voltage that all of these parts are specified at is a5v and the supply voltage is a5v for the ADC10461 and ADC10662 and g5v for the ADC12441 and ADC12451 The only other test conditions included here are the input frequencies and signal amplitudes Additional test conditions such as the ambient temperature can be found in the data sheets Only the sampling ADCs whose dynamic specifications are tested and guaranteed are included here1 2 Additional sampling ADCs made by National are the ADC0820 ADC1061 ADC1241 ADC1251 and the ADC08031 ADC08061 ADC08131 ADC08161 ADC08231 ADC10061 and ADC1031 families Dynamic Specifications for Selected Sampling ADCs Spec ADC10461 (Note 1) ADC10662 (Note 2) ADC12441 ADC12451 Resolution 10 10 12 a sign 12 a sign Conversion Time 900 ns 466 ns 13 8 ms 7 7 ms S N Unipolar (Note 3) 58 db 58 db 71 5 db 68 7 db Bipolar (Note 4) 76 5 db 73 5 db THD Unipolar (Note 5) b60 db b60 db b75 db b73 1 db Bipolar (Note 6) b75 db b78 0 db IMD Unipolar (Note 7) b73 db b78 db Bipolar (Note 8) b74 db b78 db ENOB Unipolar (Note 9) 9 9 11 6 11 1 Bipolar (Note 10) 12 4 11 9 Peak Harmonic Uni (Note 11) b82 db b82 db or Spurious Noise Bi (Note 12) b80 db b80 db Full Power BW 20 khz 20 67 khz Sampling Rate 800 khz 1 5 MHz 55 khz 83 khz Acquisition Time 3 5 ms 3 5 ms Aperture Time 100 ns 100 ns Aperture Jitter 100 ps rms 100 ps rms Note 1 There are two and four input channel members of the ADC10461 family namely the ADC10462 and ADC10464 These products have the same dynamic specifications as the ADC10461 Note 2 The ADC10664 is a 4-input channel member of the ADC10662 family (the ADC10662 has two input channels) These two products have the same dynamic specifications Note 3 ADC10461 f IN e 50 khz V IN e 4 85 V p-p ADC10662 f IN e 50 khz V IN e 4 85 V p-p ADC12441 f IN e 20 khz V IN e 4 85 V p-p ADC12451 f IN e 20 67 khz V IN e 4 85 V p-p Note 4 ADC12441 f IN e 20 khz V IN e g4 85V ADC12451 f IN e 20 67 khz V IN e g4 85V Note 5 ADC10461 f IN e 50 khz V IN e 4 85 V p-p ADC10662 f IN e 50 khz V IN e 4 85 V p-p ADC12441 f IN e 19 688 khz V IN e 4 85 V p-p ADC12451 f IN e 20 67 khz V IN e 4 85 V p-p Note 6 ADC12441 f IN e 19 688 khz V IN e g4 85V ADC12451 f IN e 20 67 khz V IN e g4 85V Note 7 ADC12441 f IN1 e 19 375 khz f IN2 e 20 625 khz V IN e 4 85 Vp-p ADC12451 f IN1 e 19 375 khz f IN2 e 20 khz V IN e 4 85 Vp-p Note 8 ADC12441 f IN1 e 19 375 khz f IN2 e 20 625 khz V IN e g4 85V ADC12451 f IN1 e 19 375 khz f IN2 e 20 khz V IN e g4 85V Note 9 ADC10461 f IN e 50 khz V IN e 4 85 V p-p ADC10662 f IN e 50 khz V IN e 4 85 V p-p ADC12441 f IN e 20 khz V IN e 4 85 V p-p ADC12451 f IN e 20 67 khz V IN e 4 85 V p-p Note 10 ADC12441 f IN e 20 khz V IN e g4 85V ADC12451 f IN e 20 67 khz V IN e g4 85V Note 11 ADC12441 f IN e 20 khz V IN e 4 85 V p-p ADC12451 f IN e 20 khz V IN e 4 85 V p-p Note 12 ADC12441 f IN e 20 khz V IN e g4 85V ADC12451 f IN e 20 khz V IN e g4 85V 6
7
AN-769 Dynamic Specifications for Sampling A D Converters LIFE SUPPORT POLICY NATIONAL S PRODUCTS ARE NOT AUTHORIZED FOR USE AS CRITICAL COMPONENTS IN LIFE SUPPORT DEVICES OR SYSTEMS WITHOUT THE EXPRESS WRITTEN APPROVAL OF THE PRESIDENT OF NATIONAL SEMICONDUCTOR CORPORATION As used herein 1 Life support devices or systems are devices or 2 A critical component is any component of a life systems which (a) are intended for surgical implant support device or system whose failure to perform can into the body or (b) support or sustain life and whose be reasonably expected to cause the failure of the life failure to perform when properly used in accordance support device or system or to affect its safety or with instructions for use provided in the labeling can effectiveness be reasonably expected to result in a significant injury to the user National Semiconductor National Semiconductor National Semiconductor National Semiconductor National Semiconductores National Semiconductor Corporation GmbH Japan Ltd Hong Kong Ltd Do Brazil Ltda (Australia) Pty Ltd 2900 Semiconductor Drive Livry-Gargan-Str 10 Sumitomo Chemical 13th Floor Straight Block Rue Deputado Lacorda Franco Building 16 P O Box 58090 D-82256 F4urstenfeldbruck Engineering Center Ocean Centre 5 Canton Rd 120-3A Business Park Drive Santa Clara CA 95052-8090 Germany Bldg 7F Tsimshatsui Kowloon Sao Paulo-SP Monash Business Park Tel 1(800) 272-9959 Tel (81-41) 35-0 1-7-1 Nakase Mihama-Ku Hong Kong Brazil 05418-000 Nottinghill Melbourne TWX (910) 339-9240 Telex 527649 Chiba-City Tel (852) 2737-1600 Tel (55-11) 212-5066 Victoria 3168 Australia Fax (81-41) 35-1 Ciba Prefecture 261 Fax (852) 2736-9960 Telex 391-1131931 NSBR BR Tel (3) 558-9999 Tel (043) 299-2300 Fax (55-11) 212-1181 Fax (3) 558-9998 Fax (043) 299-2500 National does not assume any responsibility for use of any circuitry described no circuit patent licenses are implied and National reserves the right at any time without notice to change said circuitry and specifications